How to get resource recommendations for workloads in GITA?

The Recommendation feature in GITA allows users to obtain CPU and memory requests and limits recommendations for Kubernetes workloads (such as Deployments), based on the actual consumption observed over a given period. This helps optimize resource allocation, avoiding both waste and resource shortages.

1. Access your Cluster Interface in Gita:

   

   Image 01: Cluster Selection Panel

   

   Image 02: Cluster Interface

2. Click on the Health section on the left side of the Interface screen:

   

   Image 03: Health Section

3. Inside the Health section, click on the Recommendations tab:

   

   Image 04: Recommendations Tab

4. In the Recommendations tab, select the desired workload type for analysis:

   

   Image 05: Workload Selection

5. After selecting the workload type, select the namespace:

   

   Image 06: Namespace Selection

6. Now choose the resource name for which you want to get recommendations:

   

   Image 07: Resource Selection

7. Then choose the analysis period for the recommendations (7, 14, 21, or 28 days):

   

   Image 08: Analysis Period Selection

8. Finally, click the Get Recommendation button to generate the recommendation based on the selected criteria:

   

   Image 09: Get Recommendation Button View

9. After all steps, the recommendation will be generated:

   

   Image 10: Recommendation Loading View

   Depending on the recommendation, this may take a few minutes. Thus, the user can choose to leave the Recommendations tab and will be notified with a message as soon as the recommendation is generated. Then, just click the Go button to be directed to the generated recommendation:

   

   Image 11: Recommendation Alert and Go Button

   

   Image 12: Generated Recommendation View

Legend of the Generated Recommendation:

• Current Request / Limit: Current values configured for CPU and memory request and limit in the workload.
• Recommended Request / Limit: Values suggested by GITA for request and limit, based on the actual consumption observed in the selected period.
• Usage Max / Mean: Maximum and average CPU (in millicores) and memory (in MiB) usage observed during the analyzed period.
• Usage P95 / P99: 95th and 99th percentiles of usage, indicating that 95% or 99% of the time the usage was below these values.

This information helps you adjust your workload's resources to avoid waste or shortages, making cluster usage more efficient.